Components emulating synapse-inspired activity-dependent properties (e.g. memristors), action potential spike generation and propagation in neurons offer enhanced abilities to process spatio-temporal information, needed for many real-world sensory data. Neuron-like devices have also been used to enable activity-dependent updating of non-volatile memristor arrays for unsupervised learning in data classification. Volatile Mott memristors connected in parallel with a capacitor have been used by Williams, et al to construct leaky-integrate and fire (LIF) neurons which can then be paired to form neuristors that closely mimic the action potentials and lossless signal propagation found in biological neurons. These works highlight the values of neuron-like devices to accompany synaptic mimics in artificial neural networks. They also reveal the need for more materials that exhibit tunable switching behaviors at biologically-relevant voltages—and greater understanding of how plasticity in the volatile memristive element can be used to shape and evolve the dynamic action potential in adaptive ways.

 

Biomolecular memristors pioneered recently by our group uniquely mimic the structure, composition and biophysical properties of synaptic membranes. These modular systems consist of a capacitive lipid membrane doped with one or more of many types of voltage-dependent ion channels that operate at biologically relevant voltages (\<\|200 mV\|).  Thus, we hypothesize that LIF neurons and neuristor circuits based on our biomolecular devices, which inherently integrate the necessary capacitance and volatile memristance, can enable simpler mimics of Hodgkin-Huxley neurons and operation at lower voltages. We also expect that the modularity in composition will allow for greater control in shaping the action potential, and we expect that using devices with multiple voltage-dependent mechanisms governing conductance may contribute to activity-dependent plasticity over subsequent firings. Understanding of these new biomolecular neuristor models can help realize fewer-component ionic neuristors to be implemented in artificial biomolecular neural networks.

 

For this study, we propose to leverage our existing models of volatile memristance in biomolecular systems to model and numerically simulate the behaviors of LIF neurons and neuristors incorporating two-terminal biomolecular memory circuit elements. In this way, we aim to study the contributions of voltage-dependent membrane geometry (i.e., variable capacitance,  \~ s) and ion channel dynamics ( \~ ms) towards action potential generation and signal propagation. Because these dynamic variations in the properties of the device occur at multiple time scales, we also aim to reveal how cumulative evolutions in membrane resistance and capacitance affect action potentials over subsequent cycles. Such models of biomolecular systems will complement the existing models of LIF neurons and neuristors based on Mott memristors and reveal how the biomimetic dynamics of these volatile devices can be used to construct tunable artificial neurons for brain-inspired computing.


Shaping Action Potentials and Activity-Dependent Plasticity in Bio-Molecular Neuristors.

Division / Sponsors: 
Aerospace Division

Paper Proceedings: https://asmedigitalcollection.asme.org/SMASIS/SMASIS2020/volume/84027

SMASIS2020 - Virtual Conference

Supporting the need among researchers worldwide for additional online and multimedia content, ASME now offers the ASME Conference Video Collection. This platform will collect and provide long-term access to conference presentation videos, encompassing all technology areas within ASME's technical conference and events program. About 30,000-40,000 engineers across all mechanical engineering and related disciplines participate in ASME conferences each year. This includes authors of approximately 6,000 to 8,000 technical papers published by ASME, the majority of whom also submit video presentations for our virtual conferences. The video collection will be available as an add-on to existing product packages or as a separate subscription.